Color Graphs: An Efficient Model For Two-Dimensional Cellular Automata Linear Rules

Two-dimensional nine neighbor hood rectangular Cellular Automata rules can be modeled using many different techniques like Rule matrices, State Transition Diagrams, Boolean functions, Algebraic Normal Form etc. In this paper, a new model is introduced using color graphs to model all the 512 linear rules. The graph theoretic properties therefore studied in this paper simplifies the analysis of all linear rules in comparison with other ways of its study.

💡 Research Summary

The paper introduces a novel graph‑theoretic representation for the 512 linear rules of two‑dimensional nine‑neighbour cellular automata (CA). Each cell’s next state depends on its own state and the eight surrounding cells; the authors assign the weights 1, 2, 4, 8, 16, 32, 64, 128, 256 to these positions, so that any linear rule (realizable by XOR only) can be identified by the sum of the selected weights. Traditionally, such rules are expressed as binary (0‑1) rule matrices of size (mn) × (mn), where m × n is the size of the input binary image. Because these matrices become large even for modest images, their direct analysis is cumbersome.

The authors reinterpret each rule matrix as an adjacency matrix of a directed graph. The rows (or columns) correspond to the mn cells (vertices v₁,…,v_{mn}), and a ‘1’ entry indicates a directed edge from the row‑vertex to the column‑vertex, i.e., a dependency of the former cell on the latter. Consequently, every linear rule is represented by a directed graph that may contain self‑loops (when a cell depends on itself). The paper first establishes five “basic” graphs corresponding to the fundamental rules 1, 2, 4, 8, 16 (Theorems 3.1‑3.5). Rule 1 yields a graph of mn self‑loops; Rule 2 produces m disjoint chains each of length n; Rule 4 connects vertices along one diagonal, leaving two isolated vertices; Rule 8 links each vertex to the one directly below it; Rule 16 links each vertex to the one to its right, with the first and last vertices isolated. Rules 32, 64, 128, 256 are simply transposes of the previous four, giving the same structure with reversed edge direction.

A central contribution is the definition of a “join” operation on graphs, which mirrors the XOR addition of matrices: edge presence is treated as 1, absence as 0, and 1 + 1 = 0 (mod 2). By applying this join to any subset of the five basic graphs, the authors can construct the graph of any of the 512 linear rules. The paper provides numerous concrete examples for small grids (2 × 2, 2 × 3, 3 × 4), showing the corresponding rule matrices, the resulting graphs, and classifying them as “Directed, looped” or “Directed, simple”.

The graph‑based view offers several advantages. Visualisation becomes far more intuitive than scanning large sparse matrices. Graph‑theoretic metrics—such as degree distribution, connectivity, component count, and presence of cycles—can be read directly, facilitating classification of rule dynamics. Moreover, existing graph algorithms (BFS, DFS, matching, spanning trees) become applicable for analyzing CA evolution, detecting invariant structures, or designing efficient hardware implementations. The authors argue that many applications that already exploit graph structures (e.g., VLSI layout, network routing, pattern testing) can immediately benefit from this representation.

In the conclusion, the authors claim that representing all linear CA rules as directed graphs simplifies analysis and opens new research avenues. They conjecture that the fundamental graphs and their joins correspond to classic image transformations (translations, rotations, reflections) and suggest that the same graph models could be employed in cryptographic schemes, self‑testing circuits, and random pattern generation. Overall, the paper provides a systematic method to translate the algebraic description of 2‑D linear CA into a combinatorial object, thereby bridging cellular automata theory with graph theory and offering a more accessible toolkit for both theoretical investigation and practical application.